THE SPOT* PERSONAL THERMAL COMFORT SYSTEM A. Rabbani S. Keshav University of Waterloo BuildSys, November 2016
BASIC HVAC SYSTEM http://purelle-aircare.com/images/img_commercial-duct-system.jpg
BASIC HVAC SYSTEM Air temperature is common to a set of occupants http://purelle-aircare.com/images/img_commercial-duct-system.jpg
BASIC HVAC SYSTEM Air temperature is common to a set of occupants Entire zone must be heated/cooled even if a single occupant is present http://purelle-aircare.com/images/img_commercial-duct-system.jpg
PERSONAL ENVIRONMENTAL CONTROL Ideally, provides personalized control for each building occupant Comfort measured using Predicted Mean Vote (PMV) or a personalized version called Predicted Personal Vote (PPV) http://purelle-aircare.com/images/img_commercial-duct-system.jpg
PERSONALIZED ENVIRONMENTAL CONTROL DEVICES Bauman et al 1998
PERSONALIZED ENVIRONMENTAL CONTROL DEVICES Bauman et al 1998
PERSONALIZED ENVIRONMENTAL CONTROL DEVICES Zhang et al 2010
PERSONALIZED ENVIRONMENTAL CONTROL DEVICES Sahta et al 2014
PERSONALIZED ENVIRONMENTAL CONTROL DEVICES 5 infrared sensor: Detects users clothing surface temperature Microsoft Kinect: Detects occupancy Detects location of the user Servos: Controls the direction of the 5 infrared sensor 90 infrared sensor: Detects background radiant temperature Microcontroller: Pull data from the sensors Control the rotation angle of the servos Weatherduck sensor: Detects air temperature, humidity, air velocity
ISSUES Intrusive Expensive Onerous to use Only for heating or cooling, but not both Difficult to set up No system is suitable for immediate, large-scale practical deployment
SPOT* GOALS Low per-unit cost Low operating cost Plug-and-play deployment Legacy compatibility Ease of use Ease of user training
SPOT*
SPOT*
SPOT* Office Relay Motion Sensor AC Power Control Circuit Temperature Sensor Device App Fan Heater RPi User Secure RPC Cloud, RPi, etc Cloud, RPi, etc DB App Web App Secure RPC Secure RPC Control App DB Cloud, RPi, etc HTTPS PC, Phone, RPi, etc GUI
CONTROL Reactive control, based on Occupancy Estimated PMV
MONITORING PMV PARAMETERS Air Temperature Background Infrared Radiation Air Velocity Humidity Metabolic Rate Clothing Level Measured by sensor Measured by sensor From fan Assumed Assumed Assumed
LEARNING PPV FROM PMV Training phase SPOT* determines PMV Occupant votes periodically Affine relation learnt by linear regression
CONTROL Operations phase If heating Keep heater on and fan at full speed Until PPV >= -0.5 If cooling Try each fan speed Find the lowest one such that PPV <= 0.5
WHAT IF WE RE WRONG?
WHAT IF WE RE WRONG?
RESULTS
ESTIMATED COST
DEPLOYMENT We have deployed 50+ SPOT* systems all over our campus Deployment time is about 15 minutes, including training One-page user manual If yellow light not blinking, unplug and replug Almost zero downtime
HOW EFFECTIVE IS IT?
HOW MUCH ENERGY DOES IT USE?
SURVEY RESULTS
LIMITATIONS Assumes workers do not change workspaces Did not do A-B testing Estimate comfort when not using SPOT* as PPV when unoccupied Volume cost estimates are open to question Could not calculate energy-comfort curve comfort is not a free variable
INSIGHTS Limited sensing is enough to estimate comfort GUI allows fine tuning and override Control can be lightweight Can be incorporated into device Raspberry Pi makes an excellent controller Better than a mote Provides GUI, database, hardware control
CONCLUSION We designed and implement SPOT* It meets all our design goals Low per-unit cost Low operating cost Plug-and-play deployment Legacy compatibility Ease of use Ease of user training Free and open source hardware and software!